[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

CN101064685A - Agent allocation program, method and apparatus - Google Patents

Agent allocation program, method and apparatus Download PDF

Info

Publication number
CN101064685A
CN101064685A CNA2007100898306A CN200710089830A CN101064685A CN 101064685 A CN101064685 A CN 101064685A CN A2007100898306 A CNA2007100898306 A CN A2007100898306A CN 200710089830 A CN200710089830 A CN 200710089830A CN 101064685 A CN101064685 A CN 101064685A
Authority
CN
China
Prior art keywords
node
agency
list
proxies
data volume
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CNA2007100898306A
Other languages
Chinese (zh)
Inventor
高桥俊博
水田秀行
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
International Business Machines Corp
Original Assignee
International Business Machines Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by International Business Machines Corp filed Critical International Business Machines Corp
Publication of CN101064685A publication Critical patent/CN101064685A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46Multiprogramming arrangements
    • G06F9/50Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5083Techniques for rebalancing the load in a distributed system
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L12/00Data switching networks
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L67/00Network arrangements or protocols for supporting network services or applications
    • H04L67/01Protocols
    • H04L67/10Protocols in which an application is distributed across nodes in the network
    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04LTRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L9/00Cryptographic mechanisms or cryptographic arrangements for secret or secure communications; Network security protocols
    • H04L9/40Network security protocols

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Security & Cryptography (AREA)
  • Multi Processors (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)

Abstract

The present invention discloses an agent allocation program, a method and an apparatus to properly arrange an agent quickly and properly. The invention provides a program for allocating agent on a computer comprising a plurality of nodes which are connected with each other. In the program, the arrangement of the agents in a computer wherein a plurality of nodes each having a CPU and a memory are connected to each other is performed. In the node, the node of the computer is made to execute steps: for storing a communication amount between an agent present in the node and the node with which the agent has communicated; for determining the node M having a maximum reduction amount of the communication amount when each the agent moves; for adding the agent to an agent list storing a movement candidate to the node M; and for moving the agent of the agent list to the node M.

Description

Agent allocation program, method and equipment
Technical field
The present invention relates to be used for method, more particularly, relate to the agent allocation program, method that is used to reduce internodal offered load in the multi-node environment assignment agent.
Background technology
Now, a large amount of agencies moves comprising on the computer of node, and each node has CPU and memory, and they are by fast network (multinode computer) interconnection, to carry out the physics of complexity and the simulation of economic phenomenon.The agency is the part (object parts) of the program of uninfluenced individual action entity of expression or economic entity.
In order on multinode computer, to simulate, must be to a plurality of node assignment agent.When the agency who belongs to a node sends message or when it receives message, just produces the traffic between node to the agency on another node.Because the number affects of the data that transmit between node is to the performance of whole system, thereby a kind of method that is used to reduce the traffic of needs.
The known existing method that several routines are arranged, thereby the length that for example reduces communication path reduces the method for the data volume flow through network, and be distributed between node the application program carried out under the distributional environment with balance by the CPU of these nodes and the method for offered load.Yet, any method in these methods all can't provide a kind of can be under the situation of frequent intercommunication mutually between the agency, for example using millions of agencies to carry out in the environment of simulation, come fast and suitably distribute a large amount of agencies' methods with a spot of calculating.
[patent documentation 1] disclosed not unexamined patent application No.2004-7578
Summary of the invention
The present invention makes in order to solve above-mentioned technical problem, the objective of the invention is to:
-assignment agent fast and suitably on computers,
The traffic between the-minimizing node,
Offered load between the-minimizing node, and
-distribute the agency of frequent communication mutually to same node point.
For achieving the above object, the invention provides a kind of being used in the method that comprises assignment agent on the computer of a plurality of interconnecting nodes, each node has CPU and memory, described method comprises: at each intranodal, be stored in agency resident on the described node and and this agency carried out the quantity of the data that transmit between the nodes in communication; For each agency determines node M, the data volume that this node M causes being transmitted farthest reduces when described agency is passed to described node M; Described agency is added in the list of proxies that comprises the candidate agency that will be passed to described node M; And the agency on the list of proxies is delivered to node M.
In storage because of communicating in the step of the data volume that transmits with node, according to reside at the agency on this node and carried out communicate by letter with this agency another act on behalf of between the data volume of transmission calculate and be stored in the data volume that transmits between this agency and the node.
In the step of determining node M, determine node M at each agency, this node M causes the traffic farthest to reduce this agency when present node is delivered to another node, and reduction is for deducting the data volume that transmits at present node in the data volume that node transmitted of transmitting the destination as the agency.
In transmission step, the order of successively decreasing according to the reduction of the traffic on list of proxies is to agency ordering, and the agency who transmits the predetermined quantity of higher order the list of proxies to node M from present node.
According to the present invention, the agency is suitably distributed on multinode computer, to reduce internodal offered load.Because the internodal traffic constantly reduces, can solve neatly between the agency the traffic in time and situation about changing.
Description of drawings
Fig. 1 represents the exemplary physical configuration of a plurality of nodes;
Fig. 2 represents the exemplary hardware arrangement of a node;
Fig. 3 is illustrated in the exemplary configuration of the meshed network on the three dimensional torus;
Fig. 4 is the example of executive communication between 4 nodes;
Fig. 5 is the flow chart of proxy management program; And
Fig. 6 is illustrated in the agency in the virtual environment.
Symbol description
120 multinode computers
310 nodes
410 nodes
600 virtual environments
610 agencies
620 agencies
Embodiment
Describe in detail with reference to the accompanying drawings and be used to realize the preferred embodiments of the present invention.
Node of the present invention generally comprises two chips and a memory, as Fig. 2 110 shown in.A chip comprises a CPU and two floating point units.Memory links to each other with two chips.The effect of two chips is as follows: in some cases, a chip is responsible for carrying out calculating, another chip executive communication, and two chips all are used to calculate in other cases.The exemplary physical configuration of a plurality of nodes as shown in Figure 1.The computer card of Fig. 1 is represented the node among the present invention.16 computer cards constitute a node card, and 32 node cards constitute a rack (Rack), and 64 racks constitute a multinode computer 120.These nodes interconnect by fast network, thereby are formed on the network on the three dimensional torus, as shown in Figure 3.Node 310 is corresponding to the computer card of Fig. 1.Do not use in the present invention with three dimensional torus on node between the relevant information of distance.
The agency is the object parts that move on node, can transmit to another node from a node.Fig. 6 represents to use a plurality of agencies' simulated profile.Virtual environment 600 is based upon on the multinode computer 12.A plurality of nodes and a large amount of agency carry out the complicated physics or the simulation of economic phenomenon together.Agency 610 comprises the internal model of expression variable and data and the part (function) that is used for determining described internal model.The agency takes action according to definite result.Described action comprises that calculating, storage, message send and receives.Message is sent to other agency 620,630 or receives message from other agency 620,630.
In foundation method of the present invention, the agency of each proxy records and this agent communication (node) and the traffic.According to this information, the agency of exchange predetermined quantity between all nodes.Fig. 4 is illustrated in the example that communicates between 4 nodes.Node 410 and other node communicate.Communication between the node comprise agency with node between communicate by letter.Between millions of agencies and node, will produce the huge traffic.The traffic that reduces between agency and the node will directly cause reducing the internodal traffic.
The program of the agency's that management is so a large amount of transmission is moved on each node, and transmits suitable agency to another node.The hypervisor usage quantity is less than the resource of the quantity of the resource of being used by a large amount of agencies, simple and operation rapidly, so it is little to the whole system Effect on Performance.The processing of hypervisor can be realized by specialized hardware.
Fig. 5 represents the flow chart of proxy management program.In step 510, store the traffic between each agency and the node.In step 520, determine to transmit agency's node M to it.In step 530, this agency is added to will be in candidate's list of proxies that node M is transmitted then.At last in step 540, send the agency of the higher predetermined quantity of order in the list of proxies to node M.On each node, carry out described processing.Should be noted that the node M as agency's transmission destination is not a specific node; The agency can be delivered to any node.
Describe the flow process of handling now in detail.One group of agency represents that with A a group node represents that with C the one group of agency who belongs to node k ∈ C uses B k A represents, is acting on behalf of the data volume that transmits between i and i ' ∈ A T (i, i ') expression.
In step 510, obtain each agency on given node k and belong to the data volume S that transmits between the agency of each node and it is stored on the node k.
[formula 1]
S ( i , B k ′ ) = Σ i ′ ∈ B k ′ T ( i , i ′ )
In step 520, be passed to node k ∈ C if act on behalf of i, then the transmission data volume that will be reduced according to the following formula acquisition.
ΔR(i,B k,B k’)=S(i,b k)-S(i,b k’)
Should note having obtained to make Δ R (i, B K ', B k) maximized k Max
[formula 2]
k max = arg max k ′ ΔR ( i , B k , B k ′ )
In step 530, act on behalf of i and be added to and be delivered to node k MaxCandidate agency group in:
[formula 3]
Bpool k → k max ⋐ A
In step 540, with value Δ R (i, B k, B K ') order of successively decreasing to each node k ' ∈ C from Bpool kK 'The middle B that acts on behalf of that selects one group of predetermined quantity kK ', and send it to node k ' (corresponding to the node M in the flow chart).
Each node is carried out above-mentioned processing by the time interval, can redistribute the agency to appropriate nodes with rule.As can finding out that each node is only stored the traffic that is produced by the agency who belongs to this node, and only check that the agency who belongs to this node exchanges to select the agency by above-mentioned flow chart.Therefore, need between node, not carry out complicated negotiation according to method of the present invention, and can come executable operations apace with quite a spot of calculating.

Claims (6)

1. one kind is used in the method that comprises assignment agent on the computer of a plurality of interconnecting nodes, and each node has CPU and memory, and described method comprises:
At each intranodal,
Be stored in agency resident on the described node and and this agency carried out the quantity of the data that transmit between the nodes in communication;
For each agency determines node M, the data volume that this node M causes being transmitted farthest reduces when described agency is passed to described node M;
Described agency is added in the list of proxies that comprises the candidate agency that will be passed to described node M; And
Agency on the list of proxies is delivered to node M.
2. the method for claim 1, wherein in storage because of communicating in the step of the data volume that transmits with node, according to reside at the agency on this node and carried out communicate by letter with this agency another act on behalf of between the data volume of transmission calculate and be stored in the data volume that transmits between this agency and the node.
3. the method for claim 1, wherein in the step of determining node M, determine node M at each agency, this node M causes the traffic farthest to reduce this agency when present node is delivered to another node, and reduction is for deducting the data volume that transmits at present node in the data volume that node transmitted of transmitting the destination as the agency.
4. the method for claim 1, wherein in transmission step, the order of successively decreasing according to the reduction of the traffic on list of proxies is to agency's ordering, and the agency who transmits the predetermined quantity of higher order the list of proxies to node M from present node.
5. one kind is used at the equipment that comprises assignment agent on the computer of a plurality of interconnecting nodes, and each node has CPU and memory, and described equipment comprises:
At each intranodal,
Be used to be stored in agency resident on the described node and and this agency carried out the device of the quantity of the data that transmit between the nodes in communication;
Be used to each agency to determine the device of node M, the data volume that described node M causes being transmitted farthest reduces when described agency is passed to node M;
Be used for described agency is added to the device of the list of proxies that comprises the candidate agency that will be passed to described node M; And
Be used for the agency on the list of proxies is delivered to the device of node M.
6. one kind is used in the program that comprises assignment agent on the computer of a plurality of interconnecting nodes, and each node has CPU and memory, and described program makes each node of described computer carry out following steps:
Be stored in agency resident on the described node and and this agency carried out the quantity of the data that transmit between the nodes in communication;
For each agency determines node M, the data volume that this node M causes being transmitted farthest reduces when described agency is passed to described node M;
Described agency is added in the list of proxies that comprises the candidate agency that will be passed to described node M; And
Agency on the list of proxies is delivered to node M.
CNA2007100898306A 2006-04-27 2007-04-05 Agent allocation program, method and apparatus Pending CN101064685A (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2006123367A JP2007293761A (en) 2006-04-27 2006-04-27 Arrangement program, method and device for agent
JP2006123367 2006-04-27

Publications (1)

Publication Number Publication Date
CN101064685A true CN101064685A (en) 2007-10-31

Family

ID=38649614

Family Applications (1)

Application Number Title Priority Date Filing Date
CNA2007100898306A Pending CN101064685A (en) 2006-04-27 2007-04-05 Agent allocation program, method and apparatus

Country Status (3)

Country Link
US (1) US20070255836A1 (en)
JP (1) JP2007293761A (en)
CN (1) CN101064685A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009067842A1 (en) * 2007-11-27 2009-06-04 Zte Corporation Downlink transmission system and method for borrowing spectrum resources and channel resources from adjacent cells

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8904398B2 (en) * 2011-03-31 2014-12-02 International Business Machines Corporation Hierarchical task mapping
JP6666553B2 (en) 2016-05-10 2020-03-18 富士通株式会社 Information processing apparatus, job management method, and job management program

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6192364B1 (en) * 1998-07-24 2001-02-20 Jarg Corporation Distributed computer database system and method employing intelligent agents
US6430698B1 (en) * 1998-10-05 2002-08-06 Nortel Networks Limited Virtual distributed home agent protocol
US7602786B2 (en) * 2005-07-07 2009-10-13 Cisco Technology, Inc. Methods and apparatus for optimizing mobile VPN communications
US7558859B2 (en) * 2005-10-17 2009-07-07 Microsoft Corporation Peer-to-peer auction based data distribution

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2009067842A1 (en) * 2007-11-27 2009-06-04 Zte Corporation Downlink transmission system and method for borrowing spectrum resources and channel resources from adjacent cells

Also Published As

Publication number Publication date
US20070255836A1 (en) 2007-11-01
JP2007293761A (en) 2007-11-08

Similar Documents

Publication Publication Date Title
US20020065870A1 (en) Method and apparatus for heterogeneous distributed computation
CN1959686A (en) Methods and system to model arrangements of asynchronous interface
WO2010024027A1 (en) Virtual server system and physical server selection method
US8325761B2 (en) System and method for establishing sufficient virtual channel performance in a parallel computing network
CN101042677A (en) Memory interleaving
CN1955932A (en) Method and apparatus for performance and policy analysis in distributed computing systems
CN112100185B (en) Indexing system and method for block chain data balance load
CN113377866B (en) Load balancing method and device for virtualized database proxy service
CN103885829A (en) Virtual machine cross-data-center dynamic migration optimization method based on statistics
Samani et al. Multilayer resource-aware partitioning for fog application placement
CN101064685A (en) Agent allocation program, method and apparatus
CN114217920A (en) Job scheduling method and device, computer cluster and computer readable storage medium
CN115801896A (en) Calculation network node distribution method and device, electronic equipment and storage medium
CN1805381A (en) Multi-network element emulation test system and implementation method
CN109948803A (en) Algorithm model optimization method, device and equipment
CN103473368A (en) Virtual machine real-time migration method and system based on counting rank ordering
Julián-Moreno et al. On the feasibility of 40 gbps network data capture and retention with general purpose hardware
CN115705497B (en) Quantum computer operating system and quantum computer
CN1463125A (en) Large capacity realtime stream processing method for removing dithering in using buffer memory
Chang et al. Dynamic reordering bloom filter
CN109377510B (en) Particle tracking method and system on supercomputing cluster
CN112712167A (en) Memory access method and system supporting acceleration of multiple convolutional neural networks
CN113704681B (en) Data processing method, device and super computing system
CN118467182B (en) Memory access method, computer program product, electronic device, and medium
CN114064329B (en) Time hybrid queue fragmentation system and method based on block chain

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Open date: 20071031